Estimating Continuous Production Well Flow Rates Using Ensemble Machine Learning and Deep Learning Algorithms

Abstract

This study proposes a novel approach that combines machine learning algorithms with deep learning algorithms to improve the performance of predictive models while estimating the continuous well production flow rate using daily operational conditions. This approach is a hybrid or stacked model where predictions of several base models are combined using bagging as ensemble method. The models are trained and evaluated using dataset of operational data, including well pressure, temperature, water cut, water-oil ratio, etc., collected over a period of several years. Results demonstrate that the ensemble machine learning approach achieves high accuracy. The findings suggest that the proposed approach can provide a reliable and efficient solution for estimating well production flow rates in the oil and gas industry when well flow rate measurements are not performed continuously.